TLUG Talk: An Introduction to Clustering - Revision history

Digimer at 17:34, 6 May 2011

2011-05-06T17:34:55Z

← Older revision		Revision as of 17:34, 6 May 2011
Line 1:		Line 1:
	{{talk_header}}		{{talk_header}}

			{{warning\|1=This is an older article and should only be used for historical reference. The only up to date clustering tutorial is: [[Red Hat Cluster Service 2 Tutorial]].}}

	'''''An Introduction to Clustering'''''		'''''An Introduction to Clustering'''''

Digimer at 17:24, 21 August 2010

2010-08-21T17:24:52Z

← Older revision		Revision as of 17:24, 21 August 2010
Line 16:		Line 16:

	''GTALUG; Greater Toronto Area Linux User's Group''		''GTALUG; Greater Toronto Area Linux User's Group''

	~~Note: This paper can be downloaded as a [[PDF]] or [[ODT]] in the references section at the bottom of this page.~~

	= Welcome =		= Welcome =

Digimer: /* Creating a new PV using the DRBD Partition */

2010-08-21T04:41:53Z

Creating a new PV using the DRBD Partition

← Older revision		Revision as of 04:41, 21 August 2010
Line 1,272:		Line 1,272:
	It should return output like:		It should return output like:

	<source lang="~~bash~~">		<source lang="text">
	Physical volume "/dev/drbd0" successfully created		Physical volume "/dev/drbd0" successfully created
	</source>		</source>
Line 1,280:		Line 1,280:
	Now, on both nodes, check that the new physical volume is visible by using <span class="code">pvdisplay</span>:		Now, on both nodes, check that the new physical volume is visible by using <span class="code">pvdisplay</span>:

	<source lang="~~text~~">		<source lang="bash">
	pvdisplay		pvdisplay
	</source>		</source>
	<source lang="~~bash~~">		<source lang="text">
	--- Physical volume ---		--- Physical volume ---
	PV Name /dev/sda2		PV Name /dev/sda2

Digimer: /* LVM on top of DRBD */

2010-08-21T04:41:21Z

LVM on top of DRBD

Show changes

Digimer: /* DRBD */

2010-08-21T02:56:21Z

DRBD

Show changes

Digimer: /* Cluster First Start */

2010-08-21T01:38:28Z

Cluster First Start

← Older revision		Revision as of 01:38, 21 August 2010
Line 736:		Line 736:
	== Cluster First Start ==		== Cluster First Start ==

	If everything up until now was done right, you should be able to start your cluster for the first time. It can be useful to have a separate terminal window open with a tail watching /var/log/messages so that you can see if there are any problems.		If everything up until now was done right, you should be able to start your cluster for the first time. It can be useful to have a separate terminal window open with a tail watching <span class="code">/var/log/messages</span> so that you can see if there are any problems.

	On both nodes, in dedicated terminals, run:		On both nodes, in dedicated terminals, run:

			<source lang="bash">
	clear; tail -f -n 0 /var/log/messages		clear; tail -f -n 0 /var/log/messages
			</source>

	This next step must be run on both nodes as soon as possible. If you try to start one node and wait too long to start the other node, the first node will think there is a problem and it will fence the second node. Remember the <fence_daemon post_join_delay="60"></fence_daemon> line in cluster.conf? This is where it comes into play. The value you set it the "window" you have to start both nodes before a fence is issued. The default is 6 seconds.		This next step must be run on '''both''' nodes as soon as possible. If you try to start one node and wait too long to start the other node, the first node will think there is a problem and it will fence the second node. Remember the <span class="code"><fence_daemon post_join_delay="60"></fence_daemon></span> line in <span class="code">cluster.conf</span>? This is where it comes into play. The value you set is the "window" you have to start both nodes before a fence is issued. The default is 6 seconds.

	On both nodes, in different terminals, check that cman is indeed stopped, then start it up:		On '''both''' nodes, in different terminals, check that cman is indeed stopped, then start it up:

			<source lang="bash">
	/etc/init.d/cman status		/etc/init.d/cman status
			</source>
			<source lang="text">
	ccsd is stopped		ccsd is stopped
			</source>
			<source lang="bash">
	/etc/init.d/cman start		/etc/init.d/cman start
			</source>

	If all goes well, you should see something like this in each node's /var/log/messages file:		If all goes well, you should see something like this in each node's <span class="code">/var/log/messages</span> file:

			<source lang="text">
	May 10 20:54:32 an-node01 kernel: DLM (built Mar 17 2010 12:05:05) installed		May 10 20:54:32 an-node01 kernel: DLM (built Mar 17 2010 12:05:05) installed
	May 10 20:54:32 an-node01 kernel: GFS2 (built Mar 17 2010 12:05:47) installed		May 10 20:54:32 an-node01 kernel: GFS2 (built Mar 17 2010 12:05:47) installed
Line 852:		Line 861:
	May 10 20:54:35 an-node01 openais[11175]: [CPG ] got joinlist message from node 2		May 10 20:54:35 an-node01 openais[11175]: [CPG ] got joinlist message from node 2
	May 10 20:54:36 an-node01 ccsd[11167]: Initial status:: Quorate		May 10 20:54:36 an-node01 ccsd[11167]: Initial status:: Quorate
			</source>

	Last, configure the cman daemon to start when the node boots:		Last, configure the <span class="code">cman</span> daemon to start when the node boots:

			<source lang="bash">
	chkconfig cman on		chkconfig cman on
			</source>

	Congratulations, you have a cluster!		Congratulations, you have a cluster!
Line 863:		Line 875:
	Let's looks at a common tool used in clusters. In our case, we can use DRBD to host floating VMs, a shared GFS file system, run an iSCSI target and so on.		Let's looks at a common tool used in clusters. In our case, we can use DRBD to host floating VMs, a shared GFS file system, run an iSCSI target and so on.

	DRBD		= DRBD =

	DRBD; Distributed Replicating Block Device, can be thought of as a cluster version or RAID level 1. It takes raw partitions from different nodes and ensures that whatever is written to one node is also written to the other.		DRBD; Distributed Replicating Block Device, can be thought of as a cluster version or RAID level 1. It takes raw partitions from different nodes and ensures that whatever is written to one node is also written to the other.

Digimer: /* openais.conf */

2010-08-20T22:18:39Z

openais.conf

← Older revision		Revision as of 22:18, 20 August 2010
Line 713:		Line 713:
	}		}
	interface {		interface {
	### This is the DRBD subnet, which acts as a secondary, backup network for the totem ### protocol.		### This is the DRBD subnet, which acts as a secondary, backup network for the totem
			### protocol.
	ringnumber: 1		ringnumber: 1
	bindnetaddr: 10.0.0.0		bindnetaddr: 10.0.0.0
Line 731:		Line 732:
	mode: disabled		mode: disabled
	}		}
	</~~span~~>		</source>

	== Cluster First Start ==		== Cluster First Start ==

Digimer: /* Fencing */

2010-08-20T22:17:05Z

Fencing

Show changes

Digimer: /* Fencing */

2010-08-20T22:05:40Z

Fencing

Show changes

Digimer: /* Initial Cluster Setup */

2010-08-20T21:41:36Z

Initial Cluster Setup

← Older revision		Revision as of 21:41, 20 August 2010
Line 455:		Line 455:
	Now tell Xen to reference that script by editing <span class="code">/etc/xen/xend-config.sxp</span> and changing the <span class="code">network-script</span> argument to point to this new script:		Now tell Xen to reference that script by editing <span class="code">/etc/xen/xend-config.sxp</span> and changing the <span class="code">network-script</span> argument to point to this new script:

			<source lang="text">
	#(network-script network-bridge)		#(network-script network-bridge)
	(network-script an-network-script)		(network-script an-network-script)
			</source>

	Restart <span class="code">xend</span> and everything worked, you should now be able to run ifconfig and see that all the ethX devices have matching pethX.		Restart <span class="code">xend</span> and if everything worked, you should now be able to run <span class="code">ifconfig</span> and see that all the <span class="code">ethX</span> devices have matching <span class="code">pethX</span>.

	Fencing		= Fencing =

	"The only thing you don't know, is what you don't know."		<blockquote>"The only thing you don't know, is what you don't know."</blockquote>

	Before we can discuss fencing, we must understand its critical role in clustering.		Before we can discuss fencing, we must understand its critical role in clustering.